In order to realize more accurate 3D reconstruction of featureless and featureless objects, this paper studies the automatic splicing algorithm of 3D point cloud under multi-view sensor. Firstly, the axis data was calibrated by the binocular calibration of sensors from different perspectives, and then multiple axis data were analyzed in three-dimensional space. A point cloud splicing method based on the axis fusion of sensors from multiple perspectives was proposed, so as to calculate the optimal axis data with the minimum error. Finally, the fitted axis data was used as the axis center to splice the three-dimensional point cloud in the world coordinate system. The experimental results show that the error of the proposed method is less than 0.037 mm in the measurement range of 1.3~1.9 m for the standard ball with a diameter of 144.954 2 mm, and the reconstruction of objects without and with feature points can have a good stitching effect, and the stitching time is not limited by the size of the total point cloud. This method basically meets the requirements of good stability, fast efficiency and high precision of 3D reconstruction.